yt_dlp/extractor/tenplay.py

   1 import base64
   2 import functools
   3 import itertools
   4 from datetime import datetime
   5
   6 from .common import InfoExtractor
   7 from ..networking import HEADRequest
   8 from ..utils import int_or_none, traverse_obj, urlencode_postdata, urljoin
   9
  10
  11 class TenPlayIE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:www\.)?10play\.com\.au/(?:[^/]+/)+(?P<id>tpv\d{6}[a-z]{5})'
  13     _NETRC_MACHINE = '10play'
  14     _TESTS = [{
  15         'url': 'https://10play.com.au/neighbours/web-extras/season-39/nathan-borg-is-the-first-aussie-actor-with-a-cochlear-implant-to-join-neighbours/tpv210128qupwd',
  16         'info_dict': {
  17             'id': '6226844312001',
  18             'ext': 'mp4',
  19             'title': 'Nathan Borg Is The First Aussie Actor With A Cochlear Implant To Join Neighbours',
  20             'alt_title': 'Nathan Borg Is The First Aussie Actor With A Cochlear Implant To Join Neighbours',
  21             'description': 'md5:a02d0199c901c2dd4c796f1e7dd0de43',
  22             'duration': 186,
  23             'season': 39,
  24             'series': 'Neighbours',
  25             'thumbnail': r're:https://.*\.jpg',
  26             'uploader': 'Channel 10',
  27             'age_limit': 15,
  28             'timestamp': 1611810000,
  29             'upload_date': '20210128',
  30             'uploader_id': '2199827728001',
  31         },
  32         'params': {
  33             'skip_download': True,
  34         },
  35         'skip': 'Only available in Australia',
  36     }, {
  37         'url': 'https://10play.com.au/todd-sampsons-body-hack/episodes/season-4/episode-7/tpv200921kvngh',
  38         'info_dict': {
  39             'id': '6192880312001',
  40             'ext': 'mp4',
  41             'title': "Todd Sampson's Body Hack - S4 Ep. 2",
  42             'description': 'md5:fa278820ad90f08ea187f9458316ac74',
  43             'age_limit': 15,
  44             'timestamp': 1600770600,
  45             'upload_date': '20200922',
  46             'uploader': 'Channel 10',
  47             'uploader_id': '2199827728001'
  48         },
  49         'params': {
  50             'skip_download': True,
  51         }
  52     }, {
  53         'url': 'https://10play.com.au/how-to-stay-married/web-extras/season-1/terrys-talks-ep-1-embracing-change/tpv190915ylupc',
  54         'only_matching': True,
  55     }]
  56     _GEO_BYPASS = False
  57
  58     _AUS_AGES = {
  59         'G': 0,
  60         'PG': 15,
  61         'M': 15,
  62         'MA': 15,
  63         'MA15+': 15,
  64         'R': 18,
  65         'X': 18
  66     }
  67
  68     def _get_bearer_token(self, video_id):
  69         username, password = self._get_login_info()
  70         if username is None or password is None:
  71             self.raise_login_required('Your 10play account\'s details must be provided with --username and --password.')
  72         _timestamp = datetime.now().strftime('%Y%m%d000000')
  73         _auth_header = base64.b64encode(_timestamp.encode('ascii')).decode('ascii')
  74         data = self._download_json('https://10play.com.au/api/user/auth', video_id, 'Getting bearer token', headers={
  75             'X-Network-Ten-Auth': _auth_header,
  76         }, data=urlencode_postdata({
  77             'email': username,
  78             'password': password,
  79         }))
  80         return 'Bearer ' + data['jwt']['accessToken']
  81
  82     def _real_extract(self, url):
  83         content_id = self._match_id(url)
  84         data = self._download_json(
  85             'https://10play.com.au/api/v1/videos/' + content_id, content_id)
  86         headers = {}
  87
  88         if data.get('memberGated') is True:
  89             _token = self._get_bearer_token(content_id)
  90             headers = {'Authorization': _token}
  91
  92         _video_url = self._download_json(
  93             data.get('playbackApiEndpoint'), content_id, 'Downloading video JSON',
  94             headers=headers).get('source')
  95         m3u8_url = self._request_webpage(HEADRequest(
  96             _video_url), content_id).url
  97         if '10play-not-in-oz' in m3u8_url:
  98             self.raise_geo_restricted(countries=['AU'])
  99         formats = self._extract_m3u8_formats(m3u8_url, content_id, 'mp4')
 100
 101         return {
 102             'formats': formats,
 103             'subtitles': {'en': [{'url': data.get('captionUrl')}]} if data.get('captionUrl') else None,
 104             'id': data.get('altId') or content_id,
 105             'duration': data.get('duration'),
 106             'title': data.get('subtitle'),
 107             'alt_title': data.get('title'),
 108             'description': data.get('description'),
 109             'age_limit': self._AUS_AGES.get(data.get('classification')),
 110             'series': data.get('tvShow'),
 111             'season': int_or_none(data.get('season')),
 112             'episode_number': int_or_none(data.get('episode')),
 113             'timestamp': data.get('published'),
 114             'thumbnail': data.get('imageUrl'),
 115             'uploader': 'Channel 10',
 116             'uploader_id': '2199827728001',
 117         }
 118
 119
 120 class TenPlaySeasonIE(InfoExtractor):
 121     _VALID_URL = r'https?://(?:www\.)?10play\.com\.au/(?P<show>[^/?#]+)/episodes/(?P<season>[^/?#]+)/?(?:$|[?#])'
 122     _TESTS = [{
 123         'url': 'https://10play.com.au/masterchef/episodes/season-14',
 124         'info_dict': {
 125             'title': 'Season 14',
 126             'id': 'MjMyOTIy',
 127         },
 128         'playlist_mincount': 64,
 129     }, {
 130         'url': 'https://10play.com.au/the-bold-and-the-beautiful-fast-tracked/episodes/season-2022',
 131         'info_dict': {
 132             'title': 'Season 2022',
 133             'id': 'Mjc0OTIw',
 134         },
 135         'playlist_mincount': 256,
 136     }]
 137
 138     def _entries(self, load_more_url, display_id=None):
 139         skip_ids = []
 140         for page in itertools.count(1):
 141             episodes_carousel = self._download_json(
 142                 load_more_url, display_id, query={'skipIds[]': skip_ids},
 143                 note=f'Fetching episodes page {page}')
 144
 145             episodes_chunk = episodes_carousel['items']
 146             skip_ids.extend(ep['id'] for ep in episodes_chunk)
 147
 148             for ep in episodes_chunk:
 149                 yield ep['cardLink']
 150             if not episodes_carousel['hasMore']:
 151                 break
 152
 153     def _real_extract(self, url):
 154         show, season = self._match_valid_url(url).group('show', 'season')
 155         season_info = self._download_json(
 156             f'https://10play.com.au/api/shows/{show}/episodes/{season}', f'{show}/{season}')
 157
 158         episodes_carousel = traverse_obj(season_info, (
 159             'content', 0, 'components', (
 160                 lambda _, v: v['title'].lower() == 'episodes',
 161                 (..., {dict}),
 162             )), get_all=False) or {}
 163
 164         playlist_id = episodes_carousel['tpId']
 165
 166         return self.playlist_from_matches(
 167             self._entries(urljoin(url, episodes_carousel['loadMoreUrl']), playlist_id),
 168             playlist_id, traverse_obj(season_info, ('content', 0, 'title', {str})),
 169             getter=functools.partial(urljoin, url))